🔌
New: connect Claude & other AIs to GenAIList over MCP — research the catalog and contribute to the shared knowledge base. Learn how →
// RUN — FITS
Can the NVIDIA H200 run MT-DNN?
Yes — here's how fast and up to what context, with real measured numbers.
MT-DNN on the NVIDIA H200
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 1946k | 5309.1-6327.3 t/s | ~44.2-52.7 (slow) |
| Q8 | yes | 1951k | 10618.2-12654.5 t/s | ~88.5-105.5 (slow) |
| Q4 | yes | 1954k | 18961-22597.4 t/s | ~158-188.3 (slow) |
Estimate (memory-bound). Beyond "max context" the KV cache spills to system RAM offload (depends on your system RAM) and speed drops to the "beyond" figure.
FAQ
Can the NVIDIA H200 run MT-DNN?▶
Yes. At Q4 it generates ~18961-22597.4 tok/s and fits up to 1954k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~158-188.3 tok/s.
How much context fits?▶
At Q4, up to about 1954k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.